Ternary Deep Eutectic Solvents Tasked for Carbon Dioxide Capture

Jul 18, 2014 - Department of Chemistry, Savannah State University, Savannah, Georgia ... assumes great importance.1−4 The current state-of-the art f...
1 downloads 0 Views 945KB Size
Research Article pubs.acs.org/journal/ascecg

Ternary Deep Eutectic Solvents Tasked for Carbon Dioxide Capture Leonhard L. Sze,† Shubha Pandey,†,‡ Sudhir Ravula,‡ Siddharth Pandey,‡,∥ Hua Zhao,§ Gary A. Baker,*,‡ and Sheila N. Baker*,† †

Department Department § Department ∥ Department ‡

of of of of

Chemical Engineering, University of Missouri-Columbia, Columbia, Missouri 65211, United States Chemistry, University of Missouri-Columbia, Columbia, Missouri 65211, United States Chemistry, Savannah State University, Savannah, Georgia 31404, United States Chemistry, Indian Institute of Technology, Delhi, India

S Supporting Information *

ABSTRACT: Task-specific ternary deep eutectic solvent (DES) systems comprising choline chloride, glycerol, and one of three different superbases were investigated for their ability to capture and release carbon dioxide on demand. The highest-performing systems were found to capture CO2 at a capacity of ∼10% by weight, equivalent to 2.3−2.4 mmol of CO2 captured per gram of DES sorbent. Of the superbases studied, 1,5-diazabicyclo[4.3.0]non-5-ene (DBN) gave the best overall performance in terms of CO2 capture capacity, facility of release, and low sorbent cost. Interestingly, we found that only a fraction of the theoretical CO2 capture potential of the system was utilized, offering potential pathways forward for further design and optimization of superbasederived DES systems for further improved reversible CO2 sequestration. Finally, the shear rate-dependent viscosities indicate non-Newtonian behavior which, when coupled to the competitive CO2 capture performance of these task-specific DESs despite a 1 to 2 orders of magnitude higher viscosity, suggest that the Stokes−Einstein−Debye relation may not be a valid predictor of performance for these structurally and dynamically complex fluids. KEYWORDS: Carbon dioxide capture, Deep eutectic solvent, DES, CO2 sequestration, 1,5-Diazabicyclo[4.3.0]-non-5-ene (DBN)



INTRODUCTION Carbon dioxide (CO2) emissions due to anthropogenic activities have continually increased over the past century, gaining worldwide attention due to the negative climate effects of this notorious greenhouse gas. Consequently, developing technologies to efficiently and inexpensively capture CO2 assumes great importance.1−4 The current state-of-the art for CO2 capture in post-combustion power plants is based on the absorption of CO2 by aqueous monoethanolamine (MEA) solution. However, this technology is expensive to implement and maintain, resulting in high energy penalties for sorbent regeneration. Moreover, amine-based reagents like MEA are corrosive and somewhat volatile, posing a potential threat to humans, animals, and the environment.5 In recent years, different methods to improve upon the use of volatile amines have been developed. One such methodology is based upon using ionic liquids (ILs)organic salts with melting points below 100 °Cfor reversible CO2 capture.4,6−11 ILs, due to their attractive inherent properties, including low volatility, high thermal stability, nonflammability, and excellent CO2 solubility, offer several specific advantages over aqueous MEA solutions for carbon capture. Significantly, in contrast with MEA, ILs are nonvolatile and thus do not require condensation during the regeneration (CO2 release) step. ILs can be designed to serve as physical sorbents for both CO2 and © XXXX American Chemical Society

SO2, reducing the heat requirement for regeneration and allowing for an SO2 polishing step.12 ILs are particularly well known as physical sorbents for CO2, with a high CO2 solubility that may be tuned through appropriate IL cation and anion choice.13−18 Given their tunability, “task-specific” ionic liquids (TSILs) can also be designed to chemically react with CO2 by reversibly binding CO2 through an appended amine group to form a carbamate salt,19 for example. Another strategy for the covalent capture of CO2 involves TSILs sporting alcohol functional groups that can be used in conjunction with superbases for reversible CO2 capture.20 In previous work, a mixture of the TSIL 1-(2-hydroxyethyl)-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([Im21OH][Tf2N]) and the superbase 1,8-diazabicyclo[5.4.0.]undec-7-ene (DBU) in a 1:1 molar ratio proved capable of rapidly capturing 1.04 mol of CO2 per mole of DBU at 20 °C, significantly exceeding the typical CO2 capture capacity of traditional carbamate-forming ILs. Furthermore, this system remained liquid after CO2 capture, was reversible with nearly quantitative regeneration, and was stable over several iterative capture/regeneration cycles. IL systems, in general, have been shown to capture Received: March 12, 2014 Revised: June 19, 2014

A

dx.doi.org/10.1021/sc5001594 | ACS Sustainable Chem. Eng. XXXX, XXX, XXX−XXX

ACS Sustainable Chemistry & Engineering

Research Article

of intermolecular and intramolecular interactions, as well as to what extent CO2 solubility can be tuned within a DES. Despite recent efforts investigating CO2 solubility within DESs, we are aware of no prior study reporting on a taskspecific DES system specifically designed for the chemical capture (chemisorption) of CO2. Toward this, we report on the first task-specific DES system capable of chemically binding CO2 reversibly and with high capacity. Our DES-based system consists of choline chloride (ChCl), glycerol (Gly), and a superbase (SB) component. The SB serves to deprotonate the alcohol groups of ChCl and Gly, resulting in putative alkoxide anions capable of reacting with CO2 to reversibly form alkylcarbonate salts under ambient conditions. The generalized reaction scheme summarizing this CO2 capture strategy is depicted in Figure 1. The effects of employing different SBs and

0.07−1.13 mol of CO2 per mole of IL, which equates to 19− 112 mg of CO2 per gram of IL.7,8 However, there remain several drawbacks to employing ILs for carbon capture, including the complexity and cost of their synthesis on a large scale, frequent use of unsustainable (e.g., fluorinated) components or those not vetted for environmental compatibility, and solvent byproduct wastes generated during their synthesis and purification. While numerous ILs are now available commercially, specialized TSILs best suited for CO2 separation are typically synthesized in small batches within individual academic laboratories.4 Besides, there are growing environmental concerns regarding the toxicity and environmental effects of many ILs.21,22 An emergent class of solvents related to ILs, deep eutectic solvents (DESs) share many of their favorable characteristics, such as low volatility, nonflammability, and high thermal stability.23−31 DESs are typically formed from two to three components capable of intermolecular interactions, particularly through hydrogen bonding. In the proper ratio, these components may form a eutectic with a depressed freezing point that lies well below that of the individual components. To describe this new solvent type, Abbot et al. recently coined the term deep eutectic solvent and divided DESs into four distinct types. DES types 1 to 3 consist of a quaternary ammonium halide complexed with a metal chloride, a metal chloride hydrate, and a hydrogen bond donor (such as an alcohol), respectively.32 A type 4 DES consists of a metal salt combined with a hydrogen bond donor.33 Unlike conventional IL preparation, the synthesis of DESs is very simple and straightforward, typically involving only mixing under mild heating. In fact, DES synthesis is 100% atom efficient, solventless, and requires no further purification steps, simplifying scale up. Equally important, most DESs can be made from biodegradable components whose toxicology is well characterized and may even include the use of vitamins and metabolites. Due to their many similarities with ILs, CO2 solubility within DESs has been recently studied.34−37 Zou and co-workers investigated CO2 solubility as a function of temperature (313.15−333.15 K) and pressure (0.85−12.52 MPa) within choline chloride (ChCl) + urea DESs, where the mole ratio of ChCl to urea was either 1:1.5, 1:2, or 1:2.5. The highest CO2 solubility at each temperature and pressure condition was found for the 1:2 ChCl:urea DES (the so-called reline system) and correlates with this molar ratio having the lowest melting point. In each DES studied, the CO2 solubility increased with increasing pressure and decreased with increasing temperature, suggestive of physical sorption. The dissolved CO2 mole fraction in the liquid phase ranged from a low value of 0.032 for 1:2.5 ChCl:urea at 333.15 K and 1.08 MPa to a high of 0.309 for the 1:2 ChCl:urea DES system at 313.15 K and 12.5 MPa. We note that these CO2 solubilities fall well below the CO2 solubility for imidazolium-based ILs under similar conditions, although they are on the order of the CO2 solubility observed for ammonium-based ILs.13 Francisco et al. reported CO2 solubilities even lower for the 2:1 lactic acid:ChCl DES at similar temperatures and pressures.34 In contrast, Leron and Li studied CO2 solubility within ChCl:glycerol and ChCl:ethylene glycol DESs and reported encouraging CO2 solubilities comparable to imidazolium ILs.35,36 In all of these studies, CO2 absorption was predominantly physical in nature, and additional studies are needed to determine how CO 2 absorption depends on the free volume and/or the strength

Figure 1. (A) Structures of the DES components used in this study: choline chloride, glycerol, and various superbases considered in our investigation. (B) Proposed reaction scheme for superbase-promoted CO2 capture using a task-specific DES. In this scheme, choline chloride and the superbase DBN are shown for illustration; however, a similar reaction is believed to be operative for the other superbases as well as for CO2 binding to the multiple alcohol functionalities of glycerol. Notably, the chemically captured CO2 can be facilely released upon gentle heating at 60 °C under nitrogen.

the impact of DES component mole fractions on CO2 capture capacity, absorption rate, reversibility, and the effectiveness of the system over many cycles were evaluated and are discussed herein.



EXPERIMENTAL METHOD

Synthesis of DES Systems. Choline chloride (ChCl), glycerol (Gly), 1,5-diazabicyclo[4.3.0]-non-5-ene (DBN), 7-methyl-1,5,7triazabicyclo[4.4.0]dec-5-ene (MTBD), 1,8-diazabicyclo[5.4.0]undec7-ene (DBU), 1,4-diazabicyclo[2.2.2]octane (DABCO), 1,5,7-triazabicyclo[4.4.0]-dec-5-ene (TBD), and sodium hydroxide were purchased from Sigma-Aldrich (St. Louis, MO) and used as received. In order to prepare our ternary DES mixtures, a binary mixture of ChCl and Gly was first prepared by combining ChCl and Gly in the appropriate molar ratio under vigorous stirring at 80 °C. Stirring was continued for 1 h before allowing the mixture to cool naturally to room temperature. The resulting solutions were clear and homogeneous. The binary ChCl:Gly DES mixture was further dried overnight at 50 °C under vacuum. The ChCl:Gly:SB ternary DES was subsequently prepared by heating the dried ChCl:Gly binary DES to 50 °C followed by addition of the appropriate mass of the desired SB while stirring. The final mixture (a ternary DES) was then stirred for an additional 30 min after which it was allowed to cool to ambient temperature. At this point, the DES was ready for use. It is important to note that no purification step

B

dx.doi.org/10.1021/sc5001594 | ACS Sustainable Chem. Eng. XXXX, XXX, XXX−XXX

ACS Sustainable Chemistry & Engineering

Research Article

was required and no solvents were employed in the preparation of this ternary ChCl:Gly:SB DES. CO2 Capture and Release Experiments. CO2 capture experiments were performed under ambient conditions by bubbling CO2 (99.9% grade, Airgas) at a rate of 50 mL min−1 using a mass flow controller (Aalborg Instruments and Controls, Inc.) into a known initial mass of a ternary DES while stirring. The weight of the DES was monitored over time to determine the mass of CO2 captured. After CO2 capture, the reversibility of the process was investigated by heating the ternary DES system containing chemically bound (“captured”) CO2 to 60 °C and bubbling nitrogen (ultrahigh purity, Airgas) into the mixture at a rate of 50 mL min−1 using a mass flow controller; during the desorption step, the system was open to the atmosphere. The mass loss for the sample was monitored over time corresponding to the release of CO2. Characterization Experiments. To characterize and confirm CO2 capture, 13C nuclear magnetic resonance (NMR) spectroscopy was performed using a Bruker ARX-300 Ultrashield NMR spectrometer. Fourier-transform infrared spectroscopy (FTIR) measurements were also performed to track CO2 binding and release. All FTIR measurements were performed either on a Nicolet 4700 equipped with a Thermo Smart Performer germanium crystal attenuated total reflectance (ATR) attachment or on a Nicolet Nexus 670. Viscosity measurements were performed using a Brookfield DV III Ultra cone/plate Rheometer (spindle #42) with the temperature controlled by a Fisher Isotemp circulating bath.

capture. For these experiments, a molar ratio of 1:2:6 ChCl:Gly:SB was always used to form the ternary DES mixture. As Figure 2 and Table 1 reveal, each of these systems

Figure 2. Amount of CO2 captured per gram of DES measured over time for three different task-specific DES systems comprising 1:2:6 ChCl:Gly:SB (SB = DBN, MTBD, or DBU).

proved capable of capturing CO2. After 35 min or so, the amounts of additional CO2 capture became insignificant. The DBN-containing DES performed the best overall on a mass basis (i.e., mg of CO2 captured per gram of DES), but at the 35 min mark, the performance for the MTBD-based DES was essentially equivalent. For the DBN-based DES, 103 mg of CO2 was captured per gram of DES after 35 min compared with 100 mg and 35.5 mg of CO2 capture for the MTBD- and DBUbased DESs, respectively. Given the comparable gravimetric performance, however, more CO2 was captured per mole of SB and per mole of OH groups for the MTBD-containing DES (Table 1). Using these criteria, the superbase MTBD gave the best performance for CO2 capture. In comparison to the other superbases, DBU had a much poorer CO2 capture by all performance criteria. The variation in performance for the different SBs may be explained on the basis of two main factors: basicity and sterics. Regarding the basicity, the MeCNpKBH+ values are 25.44, 24.33, and 23.89 for MTBD, DBU, and DBN, respectively.38 So, MTBD should prove only slightly more effective in deprotonating the alcohol groups of ChCl and Gly. This is manifested in the marginally higher CO2 capture ability per mole of OH observed when using MTBD over DBN. Following this argument, however, one would expect DBU to outperform DBN on an OH molar basis. Clearly, this is not the case, however. In light of this, we suggest that the decrease in CO2 capture for DBU compared with DBN arises from DBU’s larger bicyclic ring structure (Figure 1), which may sterically interfere with its ability to deprotonate OH groups in close proximity to each other (i.e., the vicinal alcohol groups of glycerol). As CO2 was added to each ternary DES, changes in the appearance of the various systems were observed. For example, during CO2 addition, the originally transparent DES mixtures generally became cloudy to opaque white (Figure S3, Supporting Information). For all ternary DES systems, the viscosity also increased over time as more CO2 was captured. It is anticipated that such a viscosity increase will negatively affect mass transfer and capture kinetics, hindering further CO2 capture. Within 6 min of CO2 exposure, the MTBD-containing DES was noticeably more viscous, continually becoming more viscous with continued CO2 capture and eventually forming a



RESULTS AND DISCUSSION For our ternary DES systems, we investigated DBN, MTBD, and DBU as prospective superbases in conjunction with a mixture of ChCl and Gly (chemical structures for these species are provided in Figure 1A). The SB in these systems serves to deprotonate the alcohol (hereafter, simply written as “OH”) group of ChCl and the multiple OH groups of Gly, which serve as handles for the subsequent chemisorption of CO2. On the basis of our previous work20 and what is known regarding superbase-promoted binding of CO2 by neutral alcohols, we propose the reaction mechanism provided in Figure 1B. We initially evaluated several additional bases for their ability to promote a ternary DES system, including DABCO, TBD, and sodium hydroxide. However, these three bases resulted in solid or highly viscous gels when combined with the binary ChCl:Gly DES and were thus not considered further for CO2 capture as such physical constraints are well known to hamper subsequent CO2 absorption.20 To verify that the observed DES mass increase upon exposure to CO2 can indeed be ascribed to CO2 capture through alkylcarbonate salt formation, 13C NMR and FTIR experiments were conducted immediately prior to and following CO2 bubbling. Comparing the 13C NMR spectra for the representative 1:2:6 ChCl:Gly:DBN system before and after CO2 bubbling reveals the appearance of a new peak near 160 ppm after CO2 exposure. This peak is fully consistent with alkylcarbonate formation after CO2 exposure (Figure S1, Supporting Information). Similarly, the ATR-FTIR spectrum collected for this DES after CO2 exposure contains new peaks in the 1600−1650 and 1300−1350 cm−1 windows, features that can be attributed to the asymmetric and symmetric stretches for unidentate-bound CO2 (Figure S2, Supporting Information). Thus, both the 13C NMR and FTIR experiments confirm the presence of chemically bound CO2 within the ternary DES after CO2 exposure. Having confirmed covalent CO2 binding within our taskspecific DES system, we turned our attention to investigating different SBs to determine which would lead to optimal CO2 C

dx.doi.org/10.1021/sc5001594 | ACS Sustainable Chem. Eng. XXXX, XXX, XXX−XXX

ACS Sustainable Chemistry & Engineering

Research Article

Table 1. Summary of CO2 Capture Capacity for Different Ternary DES Systemsa CO2 capture figures of merit superbase (SB)

DES ratio ChCl:Gly:SB

mg CO2/g DES

mmol CO2/g DES

mol CO2/mol SB

mol CO2/mol OH

OH groups/DES

CO2 bound/DES

MTBD DBU DBN DBN DBN DBN DBN

1:2:6 1:2:6 1:2:6 1:2:3 1:2:7 1:2:8 1:3:10

100 35.5 103 96 105 103 104

2.27 0.81 2.34 2.18 2.39 2.34 2.36

0.47 0.17 0.42 0.51 0.41 0.39 0.39

0.40 0.14 0.36 0.22 0.41 0.44 0.39

7 7 7 7 7 7 10

2.8 1.0 2.5 1.5 2.9 3.1 3.9

a

In practice, 35 min of CO2 flow proved sufficient to achieve equilibrium.

Figure 3. (A) Amount of CO2 captured per gram of DES as a function of time for 1:2:x molar ratio ChCl:Gly:DBN mixtures (x = 3, 6, 7, or 8). (B) Total amounts of CO2 captured per mole of DBN and per mole of alcohol (OH) group for the same DES mixtures with increasing relative DBN amounts.

gel. For the DBN-based DES, the viscosity became noticeably greater after 16 min, and upon standing for several hours, the CO2-containing mixture separated into two phases: a more viscous partially opaque white-hued liquid on the bottom and a

less viscous transparent liquid on top (Figure S3, Supporting Information). Comparing the FTIR spectra for the two distinct layers, it was evident that CO2 was present within the lower viscous phase and largely absent within the upper phase (Figure D

dx.doi.org/10.1021/sc5001594 | ACS Sustainable Chem. Eng. XXXX, XXX, XXX−XXX

ACS Sustainable Chemistry & Engineering

Research Article

capturing CO2, but the DBN superbase was utilized more efficiently at lower concentrations. Given that the superbase represents the most expensive component of this task-specific DES, this suggests that 1:2:3 ChCl:Gly:DBN is the most economical choice, while sacrificing little in terms of CO2 binding capacity. Because glycerol (Gly) contains more OH groups per molecule (i.e., three as compared to one for choline chloride, ChCl; Figure 1), we increased the relative Gly content of the ternary DES in an effort to increase CO2 binding capacity on a mass basis. Toward this, the amount of Gly relative to ChCl was increased in the DES mixture while maintaining a constant 1:1 DBN to OH ratio, using 1:2:7 and a 1:3:10 ChCl:Gly:DBN for comparison. The 1:3:10 ChCl:Gly:DBN system was found to capture CO2 at a slightly faster initial rate, but after 35 min of CO2 exposure, similar amounts of total CO2 were captured for both DES mixtures, as shown in Figure 4. The 1:2:7 and 1:3:10

S4, Supporting Information). Furthermore, the CO2 vibrational signature in the more viscous phase disappeared upon releasing the CO2 during regeneration. In contrast, the DBU-containing DES showed only a slightly noticeable increase in viscosity compared to the other two systems, an observation we assign to the much lower CO2 capture for this system. Interestingly, this mixture also separated into two phases but with most of the sample being retained in the less viscous upper phase that contained no CO2. Although the much higher viscosities observed upon CO2 binding represent an engineering challenge (and frankly, one also shared with IL systems designed for CO2 capture), there are likely ways to mitigate this, including dispersion of the DES within highly porous media and onto nanosupports.39 Another key observation we made is that the viscosity of the DES systems studied typically exhibit non-Newtonian behavior. For example, the experimental viscosities for the 1:2:6 ChCl:Gly:DBN system under ambient conditions ranged from 5450−34,613 cP, depending on the shear rate employed (Figure S5, Supporting Information). Although these viscosities are significantly (1 to 2 orders of magnitude) higher than typical IL viscosities (normally in the 50−1500 cP range),40,41 the CO2 capture performance was nonetheless competitive with and better than most known TSIL systems. For instance, the 1:1 [Im21OH][Tf2N]:DBU system discussed earlier showed a CO2 capture capacity of 7.6 wt %, equivalent to 1.72 mmol/g of sorbent.20 By comparison, despite the much higher apparent viscosities, our top DES systems showed a CO2 capture capacity of 10 wt %, corresponding to roughly 2.4 mmol of CO2 captured per gram of DES sorbent. Additional advantages include the facts that no multi-step synthesis was required (only simple mixing), no solvent wastes were produced during the DES preparation, and the system contains no halogenated species. This result also underscores the complexity of gas transport within IL/DES systems, suggesting that evoking the Stokes−Einstein−Debye (SED) relation may not be the most instructive way to predict which media will yield the greatest CO2 capture. Both the non-Newtonian behavior and SED validity represent interesting and unresolved issues that merit future inquiry. Because the DBN-containing DES gave the best performance on a gravimetric basis and represents the least expensive superbase among those studied, further examinations centered solely on the ternary ChCl:Gly:DBN DES system. Our next investigation explored how changes in the molar equivalents of the DBN component of the DES impacted CO2 capture. For these experiments, we used ternary ChCl:Gly:DBN mixtures consisting of 1:2:x molar ratios, where x equals 3, 6, 7, or 8 (i.e., 0.43, 0.86, 1.00, or 1.14 mol of DBN per OH group, respectively). As shown in Figure 3A, on a mass basis, the 1:2:7 DES captured the greatest amount of CO2 at 105 mg per gram of DES. The 1:2:6 and 1:2:8 systems behaved similarly, each capturing 103 mg of CO2 per gram of DES. We note that these values are on par with the higher levels of CO2 capture reported for TSILs8 as well as supported amine polymers.39 As anticipated, the 1:2:3 system captured slightly less: 96 mg of CO2 per gram of DES. Analyzing the moles of CO2 captured per DBN and per OH group for each DES composition elucidates the trend that as the relative molar amount of DBN increases the total moles of CO2 captured per mole of DBN decreases, while the total moles of CO2 captured per mole of OH group increases (Figure 3B and Table 1). In other words, as more DBN was added, more OH groups could be utilized for

Figure 4. Amount of CO2 captured per gram of DES over time for ternary DESs consisting of 1:2:7 and 1:3:10 molar ratios of ChCl:Gly:DBN.

ChCl:Gly:DBN mixtures also captured an equivalent 105 versus 104 mg of total CO2 per gram of DES. This amounts to 0.41 and 0.39 mol of CO2 captured per mole of OH for 1:2:7 and 1:3:10 ChCl:Gly:DBN, respectively (note that because of the 1:1 DBN to OH ratio in these systems, 0.41 and 0.39 mol of CO2 were also captured per mole of DBN for 1:2:7 and 1:3:10 ChCl:Gly:DBN as well). From these results, we can see that increasing the relative amount of Gly (relative to ChCl) made little difference and proved to be an ineffective strategy for increasing CO2 capture efficiency. The reversibility and performance during capture-and-release cycling was also tested for the representative system 1:2:6 ChCl:Gly:DBN. During each cycle, the amount of CO2 captured was maximized after 35 min of exposure to a CO2 flow of 50 mL min−1. The DES could be nearly fully regenerated within 35 min by heating to 60 °C and introducing nitrogen at the same flow rate. Over the course of three cycles, as shown in Figure 5, the CO2 capture capacity dropped from 103 to 92 to 85 mg of CO2 per gram of initial DES (i.e., 83% retention of capacity after the third cycle). We note that a similar decay was previously noted for the alcohol-appended [Im21OH][Tf2N] TSIL coupled with DBU and assigned to slight evaporative losses of the superbase component.20 Although such evaporative losses are likely to be operative here as well, this drop in capacity during cycling was also due in part to an artifact arising from the manner in which the experiment was performed. That is, in retrospect, very minor losses of DES due to sorption to the glass pipet used to introduce the gas resulted in a small drop in the initial DES mass with each cycle; we estimate this loss to be